Launching today
openviktor.com
Hire your AI employee for any role
206 followers
Hire your AI employee for any role
206 followers
Everything a great hire does. Every day. OpenViktor lives in Slack, connects to 3,000+ tools, and acts on its own. It spots problems before anyone notices and builds automations around how your company actually runs. Manages campaigns, builds apps, writes code. Runs for weeks without losing context. Not a chatbot. A coworker. Fully open source :))
I'm writing this from a hackerhouse in Barcelona ☀️
Last few months, we've been building humalike.ai, infra for Humanlike Social Agents. Last month we looked at getviktor and thought, we should make it an actual coworker with humanlike social behaviour. For now, we built a 1:1 copy of it but open source. Soon: OpenViktor X Humalike.
Today we're making it true.
OpenViktor lives in your Slack, connects to 3,000+ tools, and does the actual work: reports, code, web apps, ad campaigns.
It watches how your team works, spots problems before anyone notices, and builds automations around how your company actually runs. Nobody asks it to.
We didn't build this for a fundraise or a press cycle. We built it because we think every team deserves a coworker like this (paying $2000/month for Junior/getviktor is crazy).
Humalike is Backed by the first investor in ElevenLabs. Built by a small team (🇪🇸 x 🇵🇱) that hasn't slept much.
Marti, co-founder.
P.S. Star us on GitHub ⭐ We read every issue :))
OrangeLabs
@mcarmonas
This is seriously impressive, turning AI into an actual “coworker” instead of just a tool is a huge shift. Curious to know, how are you thinking about trust and control here? Like, how much autonomy does OpenViktor have before a human steps in or approves actions?
openviktor.com
@mcarmonas @mk_orangelabs Hi Manish! For destructive actions agent always requires permission from user. It's enforced on the architecture level, when agent decides to do something with serious consequences (like deleting a file from your google drive) it can't, until you click "approve" button that appears in slack.
It gives you visibility and full control over what agent is allowed to do, without being annoying and requiring user approval all the time:)
@mk_orangelabs Thanks for the support maestro!
@mcarmonas, thanks for creating it! Looking forward to integrating it in my Slack workflow!
And congrats to everyone who worked on it.
@eleftheria_batsou Thank you so much Eleftheria :)))))
ClawSecure
@mcarmonas The "not a chatbot, a coworker" framing is spot on. That's the mental shift most teams still haven't made. They're using AI as a search box when the real value is an agent that lives in the workflow and acts without being prompted every time.
The 3,000+ tool integrations inside Slack is what makes this practical. Most AI agent products require you to change how your team works. This one meets you where you already are. Running for weeks without losing context is the part that matters most for real production use. Context window limitations are the biggest bottleneck in agent workflows right now, and solving that unlocks use cases that short-memory agents simply can't handle.
Open source on top of all that is a strong move. Congrats on the launch!
@jdsalbego 💪. Thx Salbego!
ClawSecure
@mcarmonas my pleasure!
RiteKit Company Logo API
I'd love to try this; signup page is down, just a blank white page.
@osakasaul Fixing aasap!
@osakasaul Fixed, try now!
RiteKit Company Logo API
@mcarmonas No, not fixed.
openviktor.com
@mcarmonas @osakasaul Sorry Saul, we fixed all links on the page, but if you had /login page already loaded refreshing didn't fix it. We just made a fix for this as well, you can just refresh and connect!
Open-sourcing a full Slack-native AI coworker that connects to 3,000+ tools and proactively spots problems without being asked is a bold move against $2,000/month alternatives — the proactive behavior of observing team workflows and building automations autonomously is what separates an actual digital coworker from just another chatbot sitting in a channel. Running for weeks without losing context is a key technical claim; what's the underlying memory architecture that enables that persistence — is it a combination of vector storage and structured state, or something more custom built into the Humalike infrastructure?
openviktor.com
@svyat_dvoretski Hi! Actually it's super simple but powerful at the same time. Most of the memory is handled in the filesystem - agent creates learning files, skill files, reflection files. He has memory of organization and personal memory of your coworkers. It's stored as markdown files that agent itself creates, modifies, and looks up when needed. It's similar to how humans operate if you think about it - just taking notes. Hope this helps:))
@svyat_dvoretski AI? :)))
Congratulations on the launch!
It's hard to tell the difference between OpenVictor and Junior. Could you elaborate on what the $2k/month gets you?
Your https://openviktor.com/use-cases page is blank btw
@hex_miller_bakewell Ouops, fixing asap, thx! Junior is a pixel more complex, they join calls & have decent browser use. It's on the roadmap to do that (in 24h it's there 100%.)
Wow, seems like quite a heavy application, curious how you’re planning to handle long-term sustainability and resource efficiency.
@ravi_singh59 Most of these paid tools, charge you a lot of $ for "credits"... using OpenViktor (self-hosted) is pretty cheap :)) Thanks for the support Ravi!!
Pushary
@aadilghani Thaanks Aadil :))
@mehmet_kerem_mutlu 100%! Give it a try, have fun with it 💪